Pathogenic variants for Mendelian and complex traits in exomes of 6,517 European and African Americans: implications for the return of incidental results

Abstract

Exome sequencing (ES) is rapidly being deployed for use in clinical settings despite limited empirical data about the number and types of incidental results (with potential clinical utility) that could be offered for return to an individual. We analyzed deidentified ES data from 6,517 participants (2,204 African Americans and 4,313 European Americans) from the National Heart, Lung, and Blood Institute Exome Sequencing Project. We characterized the frequencies of pathogenic alleles in genes underlying Mendelian conditions commonly assessed by newborn-screening (NBS, n = 39) programs, genes associated with age-related macular degeneration (ARMD, n = 17), and genes known to influence drug response (PGx, n = 14). From these 70 genes, we identified 10,789 variants and curated them by manual review of OMIM, HGMD, locus-specific databases, or primary literature to a total of 399 validated pathogenic variants. The mean number of risk alleles per individual was 15.3. Every individual had at least five known PGx alleles, 99% of individuals had at least one ARMD risk allele, and 45% of individuals were carriers for at least one pathogenic NBS allele. The carrier burden for severe recessive childhood disorders was 0.57. Our results demonstrate that risk alleles of potential clinical utility for both Mendelian and complex traits are detectable in every individual. These findings highlight the necessity of developing guidelines and policies that consider the return of results to all individuals and underscore the need to develop innovative approaches and tools that enable individuals to exercise their choice about the return of incidental results.

Figure 1

Extent to which the Variant-Curation Process Enriched the Risk-Variant Set with Those that Most Likely Confer Disease Risk Violin plots showing the distribution of metrics of conservation (A), pathogenicity (B), and deleteriousness (C) between included and excluded variants in the NBS, ARMD, and PGx gene sets. (A) Mean GERP scores of variants in the risk set were significantly higher than the mean scores of the excluded variants for the NBS (p = 1.45 × 10⁻⁸⁶) and ARMD (p = 4.40 × 10⁻⁴) gene sets, but not for the PGx gene set (p = 0.066). (B) Significantly more variants were predicted to be probably damaging (blue) by PolyPhen-2 in the NBS (p = 1.99 × 10⁻¹⁹) and ARMD (p = 0.42) gene sets than in the PGx gene set (p = 0.25). The fraction of variants predicted to be benign (red) or possibly damaging (green) is denoted. (C) With the use of CADD scores, significantly more damaging variants were found in the NBS (p = 1.38 × 10⁻¹⁰⁹), ARMD (p = 9.9 × 10⁻⁴), and PGx (p = 1.47 × 10⁻⁹) risk-variant sets.

Figure 2

Comparison of Published versus Observed Carrier Estimates for NBS Conditions Published estimates of carrier frequencies were similar to estimates on the basis of a conservative manual review of NBS risk variants. Carrier estimates for sickle cell anemia (MIM 603903), beta-thalassemia (MIM 613985), and hemoglobin C disease (MIM 141900.0038) were calculated in AA individuals separately (red), and estimates for CF (MIM 219700) were calculated in AA (red) and EA (green) individuals separately, whereas all other estimates were calculated in the total ESP6500 sample (blue).

Figure 3

Plots of the Number of Pathogenic or Risk Variants per Individual AA individuals are in blue, and EA individuals are in red. Plots are shown for NBS (A), ARMD (B), and PGx (C) gene sets and a combination across gene sets (D).